Evaluation of machine learning models in predicting drought indicators (Case Study: Ajabshir area)

Document Type : Research Article


1 PhD Student, Water Resources Engineering and Management, Irrigation Department, Faculty of Agricultural Sciences and Food Industry, Islamic Azad University, Central Tehran Branch, Tehran

2 Assistant Professor, Water Resources Engineering and Management, Irrigation Department, Faculty of Agricultural Sciences and Food Industry, Islamic Azad University, Central Tehran Branch, Tehran

3 Assistant Professor, Water Resources Engineering and Management, Department of Water Science and Engineering, Faculty of Agricultural Sciences and Food Industry, Islamic Azad University, Central Tehran Branch, Tehran

4 Professor, Water Resources Engineering and Management, Water Science and Engineering, Faculty of Agricultural Sciences and Food Industry, Islamic Azad University, Central Tehran Branch, Tehran



Drought is one of the destructive phenomena with adverse impacts on water resources and water needs. Machine-learning models are among the helpful tools in time-series prediction that can provide suitable results without the requirements for basic information about a system. In this study, adaptive neuro-fuzzy inference system (ANFIS) and least square support vector regression (LSSVR) models were utilized to predict the standardized precipitation index (SPI) as a meteorological drought indicator and streamflow drought index (SDI) as a hydrological drought indicator for a period (2001-2019). Ajabshir, located in the northwest of Iran, was selected as the study area, where the data of Qaleh Chay meteorological and hydrological stations were used to calculate SPI and SDI, respectively. The precipitation and flow rate data were considered input variables of the machine-learning models in predicting the SPI and SDI, respectively. The results revealed that during the period under review, meteorological drought was more severe in 2004-2011. While in this period, hydrological drought was more severe in 2007-2011 (SPI<-3). Moreover, the prediction results of the indices showed that the performance of the LSSVR model was better than that of ANFIS for both indicators. Using LSSVR, the RMSE and MAPE error evaluation criteria for SPI were 0.74 and 0.59, respectively, while these values for SDI were obtained as 0.62 and 0.46, respectively. The findings of this study show that machine-learning models are suitable tools for predicting drought indicators. Therefore, it is suggested to use such models in predicting drought indicators in other similar regions.


Main Subjects

[1]. Tyagi S, Zhang X, Saraswat D, Sahany S, Mishra SK, Niyogi D. Flash Drought: Review of Concept, Prediction and the Potential for Machine Learning, Deep Learning Methods. Earth's Future. 2022 Nov;10(11):e2022EF002723. https://doi.org/10.1029/2022EF002723
[2]. Ray DK, Gerber JS, MacDonald GK, West PC. Climate variation explains a third of global crop yield variability. Nature communications. 2015 Jan 22; 6(1):5989. 10.1038/ncomms6989
[3]. Bahrami N. Forecasting SPI drought index using Wavelet-Neural Networks and Support Vector Regression Models (Case study: East Azerbaijan province). Master thesis. University of Tabriz. 2017; 10-120. [In Persian]
[4]. Ahmed K, Shahid S, Chung ES, Wang XJ, Harun SB. Climate change uncertainties in seasonal drought severity-area-frequency curves: Case of arid region of Pakistan. Journal of Hydrology. 2019 Mar 1;570:473-85. https://doi.org/10.1016/j.jhydrol.2019.01.019
[5]. Pande CB, Al-Ansari N, Kushwaha NL, Srivastava A, Noor R, Kumar M, et al. Forecasting of SPI and meteorological drought based on the artificial neural network and M5P model tree. Land. 2022 Nov 14;11(11):2040. 10.3390/land11112040
[6]. Mishra AK, Desai VR. Drought forecasting using stochastic models. Stochastic environmental research and risk assessment. 2005 Nov;19:326-39. 10.1007/s00477-005-0238-4
[7]. Shamshirband S, Hashemi S, Salimi H, Samadianfard S, Asadi E, Shadkani S, et al. Predicting standardized streamflow index for hydrological drought using machine learning models. Engineering Applications of Computational Fluid Mechanics. 2020 Jan 1;14(1):339-50. https://doi.org/10.1080/19942060.2020.1715844
[8]. Liu WT, Juárez RN. ENSO drought onset prediction in northeast Brazil using NDVI. International Journal of Remote Sensing. 2001 Jan 1;22(17):3483-501. https://doi.org/10.1080/01431160010006430
[9]. Han P, Wang PX, Zhang SY. Drought forecasting based on the remote sensing data using ARIMA models. Mathematical and computer modelling. 2010 Jun 1;51(11-12):1398-403. https://doi.org/10.1016/j.mcm.2009.10.031
[10]. Moreira EE, Paulo AA, Pereira LS, Mexia JT. Analysis of SPI drought class transitions using loglinear models. Journal of Hydrology. 2006 Nov 30;331(1-2):349-59. https://doi.org/10.1016/j.jhydrol.2006.05.022
[11]. Paulo AA, Pereira LS. Prediction of SPI drought class transitions using Markov chains. Water resources management. 2007 Oct;21:1813-27. https://doi.org/10.1007/s11269-006-9129-9
[12]. Rostami M, Pahlevanravi A, Moghadam Nia A. Comparison of artifial neural network models and adaptive neuro-fuzzy interface system in predicting the drought Mond basin of Fars province. Journal of Natural Enwiroment Hazard. 2016; 4(6): 21-32. [In Persian]  https://sid.ir/paper/259188/en
[13]. Nguyen V, Li Q, Nguyen L. Drought forecasting using ANFIS-a case study in drought prone area of Vietnam. Paddy and water environment. 2017 Jul;15:605-16. https://doi.org/10.1007/s10333-017-0579-x
 [14]. Mokhtarzad M, Eskandari F, Jamshidi Vanjani N, Arabasadi A. Drought forecasting by ANN, ANFIS, and SVM and comparison of the models. Environmental earth sciences. 2017 Nov;76:1-10. https://doi.org/10.1007/s12665-017-7064-0
[15]. Jalili Galgachi B. Prediction of Spi Drought Index Using Soft Calculations (Case Study of Synoptic Station of West Azerbijan Province). Master thesis. Saba college of higher education. 2017; 132. [In Persian]
[16]. Yacoub E, Tayfur G. Evaluation and assessment of meteorological drought by different methods in Trarza region, Mauritania. Water Resources Management. 2017 Feb;31:825-45. https://doi.org/10.1007/s11269-016-1510-8
[17]. Ghorbani H, Vali A, Zarepour H. Prediction and Investigation of Meteorological Drought Using SARIMA Time Series and SPI Index In Isfahan Province. jwss 2020; 23 (4) :313-328, URL: http://jstnar.iut.ac.ir/article-1-3840-fa.html [In persian]
[18 Shokrikochak S, Behnia A. Monitoring and Prediction of Khuzestan Province, Iran Drought Using SPI drought Index and Markov Chain. Irrigation Sciences and Engineering, 2013; 36(3): 1-12. [In persian]
[19]. Amiranipur M, najafzadeh M, Mohamadi S. Forecasting Drought Situation by Standardized Precipitation Evapotranspiration Index and Artificial Intelligence Models (Case Study: Kerman Synoptic Station). Journal of Meteorology and Atmospheric Science, 2021; 4(3): 260-71. [In persian] doi: 10.22034/jmas.2022.356011.1184.
[20]. Mashayekhi M, Zakeri Niri M. Meteorological, Hhydrological And Agricultural Droughts Prediction Using Wavelet Method In Tehran. Iran-Water Resources Research, 2020; 16(3): 120-32. [In persian]
[21]. Achite M, Jehanzaib M, Elshaboury N, Kim TW. Evaluation of machine learning techniques for hydrological drought modeling: A case study of the Wadi Ouahrane basin in Algeria. Water. 2022 Jan 30;14(3):431.
[22]. Nalbantis I, Tsakiris G. Assessment of hydrological drought revisited. Water resources management. 2009 Mar;23:881-97. 10.1007/s11269-008-9305-1
[23]. Jang JS. ANFIS: adaptive-network-based fuzzy inference system. IEEE transactions on systems, man, and cybernetics. 1993 May;23(3):665-85. https://doi.org/10.1109/21.256541
[24]. Milan SG, Kayhomayoon Z, Azar NA, Berndtsson R, Ramezani MR, Moghaddam HK. Using machine learning to determine acceptable levels of groundwater consumption in Iran. Sustainable Production and Consumption. 2023 Jan 1;35:388-400. https://doi.org/10.1016/j.spc.2022.11.018
[25]. Cawley GC, Talbot NL. Improved sparse least-squares support vector machines. Neurocomputing. 2002 Oct 1;48(1-4):1025-31. https://doi.org/10.1016/S0925-2312(02)00606-9
[26]. Chen LH, Chen CT, Pan YG. Groundwater level prediction using SOM-RBFN multisite model. Journal of Hydrologic Engineering. 2010 Aug;15(8):624-31. https://doi.org/10.1061/(ASCE)HE.1943-5584.0000218
[27]. Van Gestel T, Suykens JA, Lanckriet G, Lambrechts A, De Moor B, Vandewalle J. Bayesian framework for least-squares support vector machine classifiers, Gaussian processes, and kernel Fisher discriminant analysis. Neural computation. 2002 May 1;14(5):1115-47. 10.1162/089976602753633411
[28]. Samadianfard S. Prediction of SPI drought index using support vector and multiple linear regressions. Journal of Water and Soil Resources Conservation, 2017; 6(4):1-16. [In Persian] 20.1001.1.22517480.1396.
[29]. Komasi M, Malekmahmoudi M, Montaseri H. Drought forecasting by SPI and EDI indices using ANFIS method based on C-mean and SC clustering (Case study: Kohgiluyeh and Boyer Ahmad Province). Journal of Agricultural Meteorology, 2017; 5(1): 36-47. [In Persian]  doi: 10.22125/agmj.2017.54982
[30]. Goldust A,  Sobhani B. Studying drought and evaluating its prediction possibility in Ardabil province by using SPI index and ANFIS model. Geographical Research, 2015; 30(116): 135-52. [In Persian] https://jgr.ui.ac.ir/article_18102.html?lang=fa
[31]. Steinemann A. Drought indicators and triggers: a stochastic approach to evaluation 1. JAWRA Journal of the American Water Resources Association. 2003 Oct;39(5):1217-33. https://doi.org/10.1111/j.17521688.2003.tb03704.x
[32].   Arya Azar N, Ghordoyee Milan S, Kayhomayoon Z. Predicting monthly evaporation from dam reservoirs using LS-SVR and ANFIS optimized by Harris hawks optimization algorithm. Environmental Monitoring and Assessment. 2021 Nov;193:1-4.https://doi.org/10.1007/s10661-021-09495-z
[33]. Kayhomayoon Z, Arya Azar N, Ghordoyee Milan S, Berndtsson R, Najafi Marghmaleki S. Application of soft computing and evolutionary algorithms to estimate hydropower potential in multi-purpose reservoirs. Applied water science. 2023 Sep;13(9):183.https://doi.org/10.1007/s13201-023-02001-5